You don't write code. You direct it. Think of yourself as a film director — you need to know what a good scene looks like, but you don't need to operate the camera.
Vibe coding is real and it works. But not the way most people think.
The viral version of vibe coding is: you describe what you want in plain English, the AI writes it, you ship it. That's the demo. The reality is messier, slower, and requires a skill that has nothing to do with programming.
The skill is directing.
RESTAURANT: You're the executive chef. You don't stand at every station cooking every dish. You taste, you redirect, you send things back. You need to know what a good plate looks like even if you're not the one plating it. That's vibe coding — knowing what right looks like, and directing the AI until it gets there.
AI is genuinely good at: boilerplate, repetitive patterns, translating clear specifications into working code, refactoring existing code, and generating test cases. If you can describe what you want with precision, it will produce usable code more often than you'd expect.
The pattern that works: give the AI a clear brief, let it produce a first draft, review the output, then iterate. Three rounds of direction usually gets you to production-quality code.
AI fails predictably in three areas:
1. Architecture decisions. It will happily build whatever you ask for, even if the architecture is wrong. It won't tell you that your approach won't scale, that you're duplicating logic, or that you're creating technical debt. That's your job — or your strategy chat's job (Chapter 4).
2. Context drift. Long conversations lose context. The AI forgets what it said 40 messages ago. Your carefully aligned approach from the beginning of the session gets overwritten by the most recent prompt. Split long tasks into separate, focused sessions.
3. Confident errors. AI doesn't say "I don't know." It says something that sounds right but isn't. The more niche your problem, the more likely this happens. Always verify critical logic independently.
NOTE: The red flag is when the AI starts generating code that references functions or variables that don't exist in your codebase. That means it's hallucinating — inventing things. Stop, re-ground the conversation with your actual code, and start the task over with fresh context.
When code doesn't work, most people paste the error message and say "fix it." That works sometimes. Here's what works almost always:
Step 1: Paste the error message.
Step 2: Ask the AI to explain what it did and why it thought that approach would work.
Step 3: Read the explanation. Often the AI's reasoning reveals the wrong assumption, which is more useful than a blind fix.
This "explain what you did" technique catches more bugs than any other debugging pattern I've found. The AI usually knows where it went wrong once you force it to articulate its reasoning.
Vibe coding is not about knowing how to code. It's about knowing how to specify what you want, recognize when the output is wrong, and direct corrections with precision. Those are product management skills, not engineering skills.
You don't need to understand every line of code. You need to understand what the code is supposed to do and whether it's doing it. The gap between those two things is smaller than you think.